The aim of this research is to show that universal generalizations which govern the operation of primary languages hold also for non-primary language systems, including non-native languages (L2). It is hypothesized that these universals are in fact constraints on learnability, and therefore can be used as strategies for language teaching. The research focuses on second language phonology, the part of L2 acquisition responsible for the patterning of non-native speech sounds. The universal used in the proposed research is an implicational generalization about the domain in which a phonemic contrast can hold in a language. This generalization states that if a language has a phonemic contrast in morphologically composite words, it also necessarily has this contrast in morphologically simple words, but not vice versa. The claims underlying the proposed research are that this generalization holds also for L2, that it imposes constraints on the kind of language systems that humans can develop, and that it can be used as a strategy for language teaching. The strategy employed in the proposed research is to instruct learners on certain phonemic contrasts in only one of two phonological domains, one which is permitted by the generalization, and the other which is excluded by it. The prediction is that those subjects who are taught the contrast only in the phonological environment that is excluded by the generalization will generalize the learning to include the other domain, thereby learning a system that is sanctioned by the generalization, but that those learners who are trained only on the contrast that is allowed by the universal will not necessarily generalize this learning to the other environment. Two studies involving 120 subjects from four native-language backgrounds are planned. The studies address three hypotheses focusing on six different phonemic contrasts, and entail analyses of the subjects'production and perception of the contrasts in1 question. The production study will be a phonological analysis of the contrasts in both domains specified by the universal generalization;the perception study will investigate whether the implicational generalization holds also when the subjects perceive the contrasts. The health relatedness of this research is that our current pluralistic society includes many different varieties of American English, including those termed "accented" English produced by people for whom English is not a native language. These non-native forms of English are a product of 1) the rules of an individual's native language, 2) the patterns of English, and 3) principles governing how humans learn. This study will allow us to examine normal language development in second-language learners as a product of explicit training in English, especially as it relates to possible universal constraints that all humans might be subject to in all language learning. This research will lead to a better understanding of the mechanisms of normal L2 development in bilingual and limited-English populations in the United States.